Potential Bug: FMP Unresponsiveness after Receiving multiple FMP_DEV_REBOOT_TIME_REQ msg's

[MpmTwtg]

FMP does not respond after it got a valid FMP_DEV_REBOOT_TIME_REQ.
The first time it responds with an answer and after that not.

SWL2001/lbm_lib/smtc_modem_core/lorawan_packages/firmware_management_protocol/lorawan_fmp_package.c

Line 307 in ffca226

lorawan_api_send_stack_cid_req( cid_buffer, cid_request_size, stack_id );

my proposal is to remove this from line 321:

    case ANS_CMD_TASK_MASK: {
        ctx->fmp_task_ctx_mask &= ~( ANS_CMD_TASK_MASK );
        break;
    } 

And implant it with:

lorawan_api_payload_send(
                FMP_PORT, true, ctx->fmp_tx_payload_ans, ctx->fmp_tx_payload_ans_size, UNCONF_DATA_UP,
                smtc_modem_hal_get_time_in_ms( ) + smtc_modem_hal_get_random_nb_in_range( 0, ctx->fmp_ans_delay ),
                ctx->stack_id );

            ctx->fmp_tx_payload_ans_size = 0;
            ctx->fmp_task_ctx_mask &= ~( ANS_CMD_TASK_MASK );

or

status_lorawan_t status = lorawan_api_payload_send(
                FMP_PORT, true, ctx->fmp_tx_payload_ans, ctx->fmp_tx_payload_ans_size, UNCONF_DATA_UP,
                smtc_modem_hal_get_time_in_ms( ) + smtc_modem_hal_get_random_nb_in_range( 0, ctx->fmp_ans_delay ),
                ctx->stack_id );

            if(status == OKLORAWAN)
            {
                ctx->fmp_tx_payload_ans_size = 0;
                ctx->fmp_task_ctx_mask &= ~( ANS_CMD_TASK_MASK );
            }

This seems to works, would this be a correct fix and are am i missing something?

[lbm-team]

Hi MpmTwtg,
Each service is composed of 3 parts:

• on_launch(), to launch an action by the service, at this step the lorawan stack is reserved for the service, any lorawan action must be carried out in this function.
• on_update(), always called after on_launch() (or after a lorawan task if there was a lorawan task) and can be used to schedule a new on_launch() in the future (in the near future if you wish to respond to a request)
• downlink_handler() can be called at any time if a downlink is received by the lorawan stack, can also schedule a task in the future if a response is required

After received your FMP_DEV_REBOOT_TIME_REQ downlink with a valid time

can you check if the mask is correctly set with
ctx->fmp_task_ctx_mask |= ANS_CMD_TASK_MASK;

and that a new task is scheduled with fmp_add_task()

Best

[MpmTwtg]

Hi lbm-team

the ctx->fmp_task_ctx_mask |= ANS_CMD_TASK_MASK; is set
but after a second message it is reset by the fmp_add_task before it has a change to send the FMP_DEV_REBOOT_TIME_ANS

see output log
out.log

[lbm-team]

Hi MpmTwtg,

I understand your point, the modem received a new downlink that request the same answer the current answering command.

your proposition is correct to remove line 321:

case ANS_CMD_TASK_MASK: {
       // ctx->fmp_task_ctx_mask &= ~( ANS_CMD_TASK_MASK );
        break;
    } 

and to check the output of lorawan_api_send()

if( lorawan_api_payload_send(
                    FMP_PORT, true, ctx->fmp_tx_payload_ans, ctx->fmp_tx_payload_ans_size, UNCONF_DATA_UP,
                    smtc_modem_hal_get_time_in_ms( ) + smtc_modem_hal_get_random_nb_in_range( 0, ctx->fmp_ans_delay ),
                    ctx->stack_id ) == OKLORAWAN )
            {
                ctx->fmp_task_ctx_mask &= ~( ANS_CMD_TASK_MASK );
                ctx->fmp_tx_payload_ans_size = 0;
            }